# LMH6517: Question regarding channel-channel gain matching

Part Number: LMH6517

Hello all,

First the specific question then the application:

The datasheet specifies channel-channel match as +- 0.05dB.

It is not clear to me whether this holds when the gains of both channels are identical, or whether it holds over all gain combinations of both channels.

Now the application:

Our application requires measuring the relative signal levels (gain) between two inputs (say A and B) as precisely as possible.

Each of A and B is a sine wave with amplitude ranging over 3 orders of magnitude, say 1mV to 1V (but can be scaled as desired).

In spite of the above, at any particular time, signal B is never less than 0.2 * signal A.

Ideally we would like the gain between the RMS of each signal, but gain between amplitudes may also be OK.

Frequency of both A and B is always identical, and can range between 5-100 MHz (but frequency at time of measurement is unknown).

The question regarding LMH6517 arose since one idea was to use a dual VGA in order to bring both A and B to approximately the same level and then do some fine correction.

I am open to any better suggestions.

Thank you.

• Hi David,

I would do it with two precise peak detectors.

Kai

• Hi Kai,

Firstly thank you for responding. I would really appreciate some more details of your suggestion.

My background does not involve such detectors, but I understand that it would be very difficult to do precise peak detection over such a wide range of input (DR = 60dB) and frequencies (5-100 MHz).

David

• Hi David,

what is your input signal? 1mV to 1V, actually? You said that it can be scaled.

Kai

• Hi Kai,

Apparently a misunderstanding.

I meet that it could be scaled so that amplitude is say, 5mv - 5V instead of 1mv-1V. The overall dynamic range remains as is though.

Hope you can still help.

Thanks

• Hi David,

I'm not sure I'm understanding your first question. The datasheet specifies channel step accuracy of 0.05dB and also the related spec of channel-channel accuracy of 0.05dB. Are you asking if the channel-channel spec is gain dependent? Seems to me that it should apply regardless of the gain stage, but I can check on that internally. Are you trying to use the LMH6517 to normalize your signals and then account for the error after measuring the outputs? If so, how are you dynamically setting the gain based on the input amplitude?
Kai may be onto a better solution here, just let me know how I can help.

Best,

Sam

• Hi Sam and thanks for answering.

The final aim is to measure as accurately as possible the RELATIVE levels (gain) of two signals A and B.
Knowing each signals level absolute value is not really required.
My idea was to feed each of A and B into it's own VGA, and control the gain through some AGC such that the outputs of each were large enough to minimize any error in regular peak detection.
Then, knowing the control signal that was used to set the gain of each VGA, I could extract the relative levels of the two channels.
If you could think of a better solution please let me know.
Thanks
David
• Hi David,

My concern would be the AGC speed and error. I agree this could be a clever way to do it, but I'm not sure it's necessarily better.

Best,

Sam